During early development on solid substrata, Dictyostelium organize territorial waves of extracellular cAMP, with ∼6 min oscillations (McMains et al., 2008; Kuhn et al., 2021). These oscillations are recapitulated in shaking cultures of developing Dictyostelium, where extracellular cAMP is secreted, accumulated, and degraded to 5’-AMP, also with ∼6-minute repeating cycles [Supplementary Figure S1 (Kimmel, 1987; McMains et al., 2008)]. Biochemical studies show that kinases AKT and PKBR1 are phosphorylated within ∼15 s of stimulation by exogenous cAMP, followed by pERK2 at ∼30 s, and we and others have shown that the measured temporal changes in extracellular cAMP are reflected in kinase pathway activations, downstream of cAMP receptor CAR1 (Brzostowski and Kimmel, 2006; Hadwiger and Nguyen, 2011; Brzostowski et al., 2013; Liao et al., 2013). Thus, in shaking culture phospho-activations of kinases AKT, PKBR1, and ERK2 also oscillate at 6 min intervals, with pERK2 slightly delayed relative to pAKT/pPKBR1 (Liao et al., 2013). During the course of such studies, we observe that pERK1 also oscillates with identical kinetics, but slightly delayed, relative to pERK2 (Figure 2A and Supplementary Figure S2).

Others have observed that cells stimulated in culture with cAMP will activate pERK1 in delay compared to that of pERK2 and suggested that pERK1 may involve pERK2 (Hadwiger and Nguyen, 2011; Schwebs and Hadwiger, 2015; Schwebs et al., 2018; Kimmel, 2019; Nichols et al., 2019). We wished to confirm this observation and understand the connection more mechanistically. We stimulated cells with 10 μM cAMP, and although extracellular cAMP is degraded with time, a 10 μM stimulus ensures that cAMP remains elevated through >3 min (Brzostowski and Kimmel, 2006). We show maximal pERK2 at 0.5–1 min, which then rapidly declines and remains suppressed through 15 min (Figure 2B and Supplementary Figure S3). pERK1 appears with a definitive lag relative to pERK2; then pERK1 also declines. Identical data are obtained in the absence of developmental dependency (see below), which we had previously shown to be fully sensitive to activation by cAMP for cAMP production, Ca⁺² influx, Ras-GTP, pAKT, pPKBR1, and pERK2 (Liao et al., 2013).

To determine if a single dose of cAMP were sufficient to elicit these precise responses, we stimulated cells with 10 μM cAMP for 15 s, but then completely removed the stimulus, replacing cAMP with fresh DB. Identical sample times were examined as the controls. As seen, the pERK2 and pERK1 responses were identical in both cultures (Figure 2B). A single cAMP dose was sufficient to induce pERK2 activation/inactivation, followed by pERK1 activation/inactivation, essentially mimicking a single oscillation wave. Since the exogenously added cAMP was present through only 15 s, the data make it likely that ERK1 is activated and deactivated as part of a downstream cascade circuit, rather than directly via CAR1. Potentially pERK2 may be required to activate ERK1, and further, since pERK2 decline seems to follow pERK1 activation (Figure 2A), pERK1 may be required to suppress pERK2.

We directly examined the activation of ERK1 in WT cells and cells lacking the ERK2 gene. As previously suggested (Hadwiger and Nguyen, 2011; Schwebs and Hadwiger, 2015; Schwebs et al., 2018; Kimmel, 2019; Nichols et al., 2019), erk2-null cells are unable to activate pERK1 following stimulation with cAMP (Figure 3A), data consistent with a dependent role for pERK2 in ERK1 regulation. To ensure that the phenotype was not secondary to a developmental defect or even a cAMP response, we additionally looked to ERK2/ERK1 regulation in growth-phase cells that are responsive through a different chemoattractant, folate, and a different GPCR, Far1 (Figure 3B). We observe fundamentally similar time course regulations for ERK2 and ERK1 in control cells, and again with full dependency on pERK2 for ERK1 phosphorylation. We then expressed a FLAG-ERK2 fusion in erk2-null cells (Supplementary Figure S4) to further follow pERK1 dependency on pERK2. These cells show rescued cAMP-dependent activation of pERK2 and rescued activation of pERK1, which were temporally similar to WT (Supplementary Figure S4).

If pERK2 were required for the regulation of ERK1 activation, we hypothesized that maintaining persistent pERK2 levels might similarly extend pERK1 activation. We approached this in two manners.

Two phosphatases (PPases) are involved in ERK2 de-phosphorylation during early development, Protein Phosphatase 2A (PP2A) and Mpl1 (Rodriguez et al., 2008), with PP2A the more significant (Rodriguez et al. al., 2008; (Lee et al., 2008); data not shown). We, thus, focused to the regulation of pERK2 and pERK1 comparing WT to B56-null cells, which lack the essential B56 regulatory subunit of PP2A (Lee et al., 2008). In B56-null cells, ERK2 is hyper-phosphorylated and pERK2 levels are significantly extended temporally compared to WT (Figure 3C). Further, ERK1 remains phosphorylated through the entire time course, consistent with phosphorylation of ERK1 induced through pERK2. Fundamentally similar data are obtained in growing cells, using folate as the stimulus (Supplementary Figure S5A).

Next, we maintained pERK2 levels through inhibition of ERK2 de-phosphorylation (Brzostowski and Kimmel, 2006). We had previously shown that kinase and PPase activities for ERK2 were under separate regulations. The kinase is transiently activated (see Figure 1) but is then rapidly adapted to continuous cAMP stimulation. Conversely, inhibition of the PPase for ERK2 persists in the presence of continuous receptor stimulation. Here, we added 1 μM cAMP every 30 s to maintain CAR1 saturation and PPase inhibition. As seen (Figure 3D), inhibition of PPase with continued cAMP stimulation suppresses pERK2 de-phosphorylation; persistent pERK2 activation is followed with prolonged pERK1 activation. The kinetics of both pERK2 and pERK1 are extended compared to controls that had received only single cAMP dose (Figure 3D), again supporting activation of pERK1 downstream of pERK2.

We performed a parallel experiment in growing cells, using folate as the stimulus. When 1 μM folate is added every min to maintain Far1 receptor saturation, inhibition of PPase is also persistent, limiting de-phosphorylation of ERK2. Continuous pERK2 activation promotes continued pERK1 activation. Again, the kinetics of both pERK2 and pERK1 are extended compared to controls that had received only a single folate dose (see Supplementary Figure S5B).

Persistent extracellular cAMP levels can also be generated through direct inhibition of the extracellular phosphodiesterase PDE1 with 10 mM DTT (Brzostowski and Kimmel, 2006). Maintenance of a saturating extracellular cAMP stimulus with 10 mM DTT also inhibits de-phosphorylation of pERK2, which supports persistent pERK1 activation. We show that both pERK2 and pERK1 are extended in DTT-treated cells, compared to untreated cells (Supplementary Figure S5C).

With regulation of ERK1 defined downstream of ERK2, we sought to re-visit the temporal connection between ERK1 activation and ERK2 de-phosphorylation. We hypothesized that in the absence pERK1, ERK2 should not exhibit downregulation but shown substantially persistent phosphorylation. Indeed, cells deficient for ERK1, show activation of pERK2, without a rapid de-phosphorylation response to either cAMP (Figure 4A) or folate (Supplementary Figure S6).

Next, we expressed an ERK1-RFP fusion in erk1-null cells. These cells show rescued activation of pERK1 at a mobility consistent with the size of the ERK1-RFP fusion, and also pERK2 activation and rescued de-phosphorylation of pERK2 that were temporally similar to WT (Figure 4B).

Although ERK2 is argued to have characteristics similar to other ERKs that are not phosphorylated by MEKs, previous work in Dictyostelium has suggested MEK1 as the pERK1 activator (Ma et al., 1997; Sobko et al., 2002; Mendoza et al., 2005; Nguyen et al., 2010). We confirm these studies and further show that mek1-null cells, which lack pERK1, also show extended pERK2, paralleling the previous studies using erk1-nulls (Figure 4C). Expression of a GFP-MEK1 fusion in mek1-null cells (Supplementary Figure S7) showed rescued cAMP-dependent activation of pERK1 but also rescued de-phosphorylation of pERK2, which temporally parallels WT (Supplementary Figure S7).

Thus, following CAR1 activation, we observe a rapid activation of pERK2, leading to MEK1-mediated phosphorylation of pERK1, which, we suggest, feed-back inhibits pERK2, probably through activation of a PPase (Figure 4D). With de-phosphorylation of pERK2, RegA is reactivated, bringing the elevated levels of cAMP back to base line, which defines a single cAMP oscillation.

As stated previously, a major known mechanistic action of pERK2 is in control of intra- and extra-cellular levels of cAMP (Sawai et al., 2005; Brzostowski and Kimmel, 2006), where it functions to inhibit RegA, the major intracellular PDE that degrades cAMP (Sawai et al., 2005). Thus, the co-activation of ACA and pERK2 by extracellular cAMP, leads to intracellular cAMP accumulation and activation of the cAMP-dependent protein kinase (PKA). Mechanistically, this occurs through cAMP binding to the regulatory subunit of PKA (PKAreg), which promotes dissociation from and activation of the PKA catalytic subunit (PKAcat). Several models had invoked PKAcat to feedback inhibit ACA or ERK2 directly (Maeda et al., 2004), although subsequent published data do not support these conclusions (Sawai et al., 2005; Brzostowski and Kimmel, 2006). Still, possibly PKAcat might mediate pERK2-induced activation of ERK1. For this, we studied cells with diminished cAMP/PKA signaling [acaA-nulls (Hirose et al., 2021)] or with constitutively active PKA [pkaR-nulls (Shaulsky et al., 1998)].

Interestingly, pERK2 and pERK1 are activated in acaA-null cells with similar kinetics to WT (Figure 5A). Thus, a pERK2-dependent increase in cAMP/PKA is not required to drive ERK1 activation. However, pERK1 in cells lacking ACA (acaA ⁻ cells) is significantly extended relative to WT (Figure 5A), regardless that ERK2 is de-activated. Dictyostelium possess two adenylate cyclases, ACR and ACG, in addition to ACA. These other ACs do not couple with CAR1, but are regulated differently (Hirose et al., 2021). We, thus, followed pERK2/pERK1 regulation in acaA–/acrA–/acgA–null cells (Hirose et al., 2021), which lack all 3 adenylate cyclases and fail to produce any detectable cAMP (Hirose et al., 2021). Results were identical to that of acaA-null cells (Supplementary Figure S8A). In the absence of cAMP and activated PKA, pERK1 remains elevated through the entire time course. We observed a similar loss of pERK1 de-activation in folate-stimulated cells that lack either ACA or PKAcat (Supplementary Figure S8B).

Based on data above (Figure 5A, Supplementary Figure S8), where reduced cAMP leads to extended pERK1, cells with constitutively activated PKA might be expected to show reduced pERK1 signaling than WT. To this, we used cells lacking the inhibitory regulatory subunit of PKA, pkaR-null cells (Shaulsky et al., 1998). Indeed, pERK1 activation is attenuated and de-phosphorylation more rapid in pkaR-null with constitutively elevated PKA than in WT (Figure 5B).

Collectively, we suggest PKA activation directs the de-phosphorylating, inactivation loop to pERK1 (Figure 5C). We are unable to determine if PKA suppresses MEK1 activity or if PKA promotes pERK1 de-phosphorylation. Regardless, the low basal level of pERK1 observed in both acaA-null and acaA–/acrA–/acgA–null cells, which have suppressed PKA activity, is consistent with PKA-regulated reduction in pERK1 levels.

Developmental cAMP oscillations are characterized by a rise in cAMP synthesis and accumulation, followed by cessation of cAMP synthesis and cAMP degradation, with the cycle repeating at defined temporal cycles. GPCR de-sensitization to a saturating or a continuous stimulation is a well-described phenomenon, often involving homologous (or heterologous) ligand-induced receptor phosphorylation (Xiao et al., 1999; Artemenko et al., 2016). Dictyostelium CAR1 is also modified by cAMP-induced phosphorylation (Klein et al., 1985; Hereld et al., 1994) and, compared to WT, CAR1 mutants with reduced phosphorylation show extended ACA activity in response to a persistent cAMP stimulation (Brzostowski et al., 2013). There are components downstream of CAR1 [e.g., RAS inhibition through GAP activation (McMains et al., 2008; King and Insall, 2009; Charest et al., 2010; Swaney et al., 2010; Rappel and Firtel, 2012; Liao et al., 2013; Scavello et al., 2017)] that also function to suppress ACA, and others have suggested that these pathways are the more significant for ACA suppression (Cai et al., 2010; Scavello et al., 2017). However, we have shown that cells that are fully insensitive to cAMP stimulation (e.g., RAS, mTORC2 activation) are also fully responsive to activation through a different GPCR, Far1 (Liao et al., 2013; Meena and Kimmel, 2017). Therefore, we have argued that primary adaptation is localized high in the circuit involving CAR1 and its associated G proteins and that downstream inhibitory feedback are more transitory.

Thus, we explored the regulation of ERK2, and, accordingly RegA and cAMP, in response to differential GPCR ligand-induced stimulation. First, cells were stimulated with saturating 10 μM cAMP. We observed rapid activation of pERK2, followed by ERK2 de-phosphorylation and activation of pERK1. At 5 min, CAR1 is still in a desensitized state (Liao et al., 2013). Here, the culture was split. An aliquot was washed of cAMP and replaced with buffer without cAMP (DB). A separate aliquot was washed of cAMP and replaced with fresh 10 μM cAMP. Both show identical regulatory patterns for ERK2 and ERK1. ERK2 remains inactive and pERK1 retains activity until later time points, when it also declines (Figure 6A). Additionally, we activated an additional aliquot at 5 min with folate. The GPCR folate receptor is coupled to Gα4, and not Gα2 as is CAR1, but the same Gβγ. In these cells, we see a new cycle of ERK2 response and de-phosphorylation and ERK1 re-activation. (Figure 6A). In a similar experiment, we had shown cAMP adapted cells were responsive to folate for Gβγ-dependent re-activation of RAS and mTORC2 in a path toward activated ACA (Liao et al., 2013; Meena and Kimmel, 2017). Therefore, the primary adaptation for ACA must be independent of downstream pathway feed-back.

To reinitiate signal propagation, receptors must become re-sensitized (de-adapted). De-adaptation is not fully understood but involves receptor de-phosphorylation and G protein re-coupling. The individual downstream inhibitory modules certainly play critical roles for acute aspects to cAMP signaling, but they are bypassed upon receptor stimulation. Therefore, we suggest that receptor adaptation/re-sensitization is the more critical to allow oscillatory response/relay. Once all the pathways are fully re-set, a slight activation of CAR1 is sufficient to initiate another cycle of cAMP amplification and oscillation (see Figures 1, 6B).

During early development of Dictyostelium, secreted cAMP serves as a chemoattractant to organize cells at centers of aggregation (Meima and Schaap, 1999; Jaiswal et al., 2012a; Jaiswal et al., 2012b; Fukujin et al., 2016; Scavello et al., 2017; Singer et al., 2019), The receptors for cAMP (CAR1) are surface chemoattractant GPCRs, coupled to Gα2 and Gβγ subunits, which transduce the extracellular signal for directed cell movement (Janetopoulos et al., 2001; Xu et al., 2010; Tariqul Islam et al., 2018; Adhikari et al., 2021). In addition, CAR1 orchestrates a periodic activation/de-activation/re-activation pathway for cAMP generation and clearing, which relays the cAMP signal (Louis et al., 1993; Insall et al., 1994b; Hereld et al., 1994; Thomason et al., 1999; McMains et al., 2008; Kriebel and Parent, 2009; Xu et al., 2010; Brzostowski et al., 2013; Cao et al., 2014). In three-dimensional space, cAMP is seen as waves, oscillating with a set period. The cAMP waves move outward from an initial center, with responsive cells migrating “up” the wave gradient and relaying the cAMP to co-ordinate more cells. To ensure inward directional movement and outward cAMP relay as the wave pass through a territory, cells go through activated and adapted states. In shaking culture, coordinated cAMP oscillations are seen through the entire culture as peaks and valleys with periodicities of ∼6 min (see Supplementary Figure S1).

We present an integrated, cross-regulation pathway model involving activating/adaptive and feed-forward/feed-back loops, for directed oscillatory cAMP signal-relay/response during the development of Dictyostelium (Figure 7). Blue lines represent activation pathways, whereas red lines are inhibitory. In the model, we highlight the most critical elements for temporally regulated activating/inhibitory oscillating circuitry. Other components with significant input roles are discussed in more detail below. Still, we recognize that we have not included every cAMP signaling component, with some having possible feed-back/feed-forward modulating influences.

During Dictyostelium development, oscillating waves of cAMP organize cells for multi-cellular aggregation. The cross regulation of ERK2 and ERK1, integrated with input to cAMP stability, explains oscillatory cAMP wave production, cAMP signal-relay, and directed chemotactic aggregation. cAMP activates ACA and pERK2. cAMP is synthesized and stabilized through inhibition of RegA. Secreted cAMP recruits additional cells for signal-relay and aggregation, but also elicits adaptation of CAR1; ACA and ERK2 kinase are deactivated. In parallel, ERK1 is activated, which inhibits pERK2. Intracellular cAMP is cleared by the re-activated RegA and other downstream pathways return to a basal state. Extracellular cAMP is degraded by secreted PDE1 and cells become re-sensitized for another pulse of extracellular cAMP stimulation.

Thus, we have proposed an integrated model that describes how periodic cellular cycling from a basally responsive state, to activation, to de-activation, and, finally, back to a basal, re-responsive state can generate the oscillating cAMP signals that are required to direct multicellular development of Dictyostelium.

Dictyostelium AX3 WT strain, erk2-null (Nichols et al., 2019), erk1-null (Nichols et al., 2019), mek1-null (Sobko et al., 2002; Mendoza et al., 2005), B56-null (Lee et al., 2008), acaA-null (Hirose et al., 2021), acaA-/acrA-/acgA-null (Hirose et al., 2021), erk1 ^-/ERK1-RFP, mek1 ^-/GFP-MEK1, pkaR-null (Shaulsky et al., 1998), pkaC-null (Mann et al., 1992), and erk2 ^-/FLAG-ERK2 cells were grown axenically in HL5 medium containing 100 μg/mL of ampicillin and 100 μg/mL streptomycin at 22°C in suspension culture, shaking at ∼180 rpm, to a density of 1–1.5 × 10⁶ cells/mL. erk2-, erk1-, mek1-, B56-, acaA-/acrA-/acgA-, and pkaC-null cells were grown under selection in 5 μg/mL blasticidin (InvivoGen # ant-bl). erk1 ^-/ERK1-RFP cells, mek1 ^-/GFP-MEK1 cells, and erk2 ^-/FLAG-ERK2 cells were grown under selection in 10 μg G418 (Gold Biotechnology # G-418-5). Cell lines are available at dictyBase (Basu et al., 2013; Fey et al., 2013).

Cells were grown to a density of 1–1.5 × 10⁶ cells/mL. For development, cells were washed twice with developmental buffer [DB (Meena and Kimmel, 2016)] at 22°C and resuspended in DB to a density of 2 × 10⁷ cells/mL. The suspension was gently shaken at 30–40 rpm at 22°C. A time-controlled peristaltic pump was used to rapidly (<20 s) add cAMP to 75 nM at 6 min intervals for 5 h. Pulsing was stopped and cells were shaken for an additional 12 min. To deactivate ACA, cells were washed and resuspended in DB containing 3 mM caffeine (freshly prepared). After 20 min, cells were washed and resuspended in DB (Meena and Kimmel, 2016). acaA-null and acaA-/acrA-/acgA-null cells were not treated with caffeine.

cAMP developed cells at 2 × 10⁷ cells/mL were shaken for 30 s, and then stimulated with a bolus of cAMP (Meena and Kimmel, 2016). To monitor the kinetics for immunoblotting (see below), a 75 μL of sample was added to 25 μL (4x) SDS lysis buffer. For Jess Protein Analysis (see below), 20 μL of the sample was mixed with 80 μL of RIPA buffer supplemented with PhosStop and Protease inhibitors.

Cells grown at a density of 1–1.5 × 10⁶ cells/mL were washed twice and resuspended in DB and shaken for 30–90 min at 22°C. Cells were then stimulated with a bolus of folate. Samples for immunoblotting were collected as above, for cAMP.

pERK1, pERK2, pPKBR1, FLAG-ERK2, GFP-MEK1, and actin were monitored in whole cell lysates by immunoblotting following gel electrophoresis (Bio-Rad, 4%–15% tris glycine gels), with antibodies against human pERK^T202/Y204 [1:1000, Cell Signalling # 9101 for pERK1 and pERK2], human p70S6K^T389 [1:1000; Cell signaling #9205 for pT470-PKBR1], RFP [1:1000; chromotek # 6G6], GFP [1:1000; Chromotek # pabg1], FLAG [1:1000; Sigma-Aldrich #F3165], and actin [Santa Cruz Biotechnology # SC-1616] (Meena and Kimmel, 2016; Jaiswal and Kimmel, 2019). We validated pERK1, pERK2, and actin immunoblots using Jess Protein Simple (Bio-Techne). A 1:50 dilution of pERKT202/Y204 and a 1:100 dilution of actin-HRP were prepared. For the experiment, 2 μg of protein was loaded into a well in the cartridge, and the remaining steps, including loading other reagents, analysis, and image processing, were carried out according to the manufacturer’s instructions. All immunoblot experiments were replicated at least 3 times, with data shown for both cAMP and folate, unless otherwise described.